/* Handling of recursive HTTP retrieving.
- Copyright (C) 1995, 1996, 1997, 2000 Free Software Foundation, Inc.
+ Copyright (C) 1995, 1996, 1997, 2000, 2001 Free Software Foundation, Inc.
This file is part of GNU Wget.
#include "fnmatch.h"
#include "host.h"
#include "hash.h"
+#include "res.h"
#ifndef errno
extern int errno;
extern char *version_string;
-#define ROBOTS_FILENAME "robots.txt"
-
static struct hash_table *dl_file_url_map;
static struct hash_table *dl_url_file_map;
/* List of HTML files downloaded in this Wget run. Used for link
- conversion after Wget is done. */
+ conversion after Wget is done. This list should only be traversed
+ in order. If you need to check whether a file has been downloaded,
+ use a hash table, e.g. dl_file_url_map. */
static slist *downloaded_html_files;
+\f
+/* Functions for maintaining the URL queue. */
-/* List of undesirable-to-load URLs. */
-static struct hash_table *undesirable_urls;
-
-/* List of forbidden locations. */
-static char **forbidden = NULL;
+struct queue_element {
+ const char *url;
+ const char *referer;
+ int depth;
+ struct queue_element *next;
+};
-/* Current recursion depth. */
-static int depth;
+struct url_queue {
+ struct queue_element *head;
+ struct queue_element *tail;
+ int count, maxcount;
+};
-/* Base directory we're recursing from (used by no_parent). */
-static char *base_dir;
+/* Create a URL queue. */
-/* The host name for which we last checked robots. */
-static char *robots_host;
+static struct url_queue *
+url_queue_new (void)
+{
+ struct url_queue *queue = xmalloc (sizeof (*queue));
+ memset (queue, '\0', sizeof (*queue));
+ return queue;
+}
-static int first_time = 1;
+/* Delete a URL queue. */
-/* Construct the robots URL. */
-static struct urlinfo *robots_url PARAMS ((const char *, const char *));
-static uerr_t retrieve_robots PARAMS ((const char *, const char *));
-static char **parse_robots PARAMS ((const char *));
-static int robots_match PARAMS ((struct urlinfo *, char **));
+static void
+url_queue_delete (struct url_queue *queue)
+{
+ xfree (queue);
+}
+/* Enqueue a URL in the queue. The queue is FIFO: the items will be
+ retrieved ("dequeued") from the queue in the order they were placed
+ into it. */
-/* Cleanup the data structures associated with recursive retrieving
- (the variables above). */
-void
-recursive_cleanup (void)
+static void
+url_enqueue (struct url_queue *queue,
+ const char *url, const char *referer, int depth)
{
- if (undesirable_urls)
- {
- string_set_free (undesirable_urls);
- undesirable_urls = NULL;
- }
- if (dl_file_url_map)
- {
- free_keys_and_values (dl_file_url_map);
- hash_table_destroy (dl_file_url_map);
- dl_file_url_map = NULL;
- }
- if (dl_url_file_map)
- {
- free_keys_and_values (dl_url_file_map);
- hash_table_destroy (dl_url_file_map);
- dl_url_file_map = NULL;
- }
- undesirable_urls = NULL;
- free_vec (forbidden);
- forbidden = NULL;
- slist_free (downloaded_html_files);
- downloaded_html_files = NULL;
- FREE_MAYBE (base_dir);
- FREE_MAYBE (robots_host);
- first_time = 1;
+ struct queue_element *qel = xmalloc (sizeof (*qel));
+ qel->url = url;
+ qel->referer = referer;
+ qel->depth = depth;
+ qel->next = NULL;
+
+ ++queue->count;
+ if (queue->count > queue->maxcount)
+ queue->maxcount = queue->count;
+
+ DEBUGP (("Enqueuing %s at depth %d\n", url, depth));
+ DEBUGP (("Queue count %d, maxcount %d.\n", queue->count, queue->maxcount));
+
+ if (queue->tail)
+ queue->tail->next = qel;
+ queue->tail = qel;
+
+ if (!queue->head)
+ queue->head = queue->tail;
}
-/* Reset FIRST_TIME to 1, so that some action can be taken in
- recursive_retrieve(). */
-void
-recursive_reset (void)
+/* Take a URL out of the queue. Return 1 if this operation succeeded,
+ or 0 if the queue is empty. */
+
+static int
+url_dequeue (struct url_queue *queue,
+ const char **url, const char **referer, int *depth)
{
- first_time = 1;
+ struct queue_element *qel = queue->head;
+
+ if (!qel)
+ return 0;
+
+ queue->head = queue->head->next;
+ if (!queue->head)
+ queue->tail = NULL;
+
+ *url = qel->url;
+ *referer = qel->referer;
+ *depth = qel->depth;
+
+ --queue->count;
+
+ DEBUGP (("Dequeuing %s at depth %d\n", qel->url, qel->depth));
+ DEBUGP (("Queue count %d, maxcount %d.\n", queue->count, queue->maxcount));
+
+ xfree (qel);
+ return 1;
}
-/* The core of recursive retrieving. Endless recursion is avoided by
- having all URLs stored to a linked list of URLs, which is checked
- before loading any URL. That way no URL can get loaded twice.
+static int descend_url_p PARAMS ((const struct urlpos *, struct url *, int,
+ struct url *, struct hash_table *));
+
+/* Retrieve a part of the web beginning with START_URL. This used to
+ be called "recursive retrieval", because the old function was
+ recursive and implemented depth-first search. retrieve_tree on the
+ other hand implements breadth-search traversal of the tree, which
+ results in much nicer ordering of downloads.
+
+ The algorithm this function uses is simple:
+
+ 1. put START_URL in the queue.
+ 2. while there are URLs in the queue:
+
+ 3. get next URL from the queue.
+ 4. download it.
+ 5. if the URL is HTML and its depth does not exceed maximum depth,
+ get the list of URLs embedded therein.
+ 6. for each of those URLs do the following:
+
+ 7. if the URL is not one of those downloaded before, and if it
+ satisfies the criteria specified by the various command-line
+ options, add it to the queue. */
- The function also supports specification of maximum recursion depth
- and a number of other goodies. */
uerr_t
-recursive_retrieve (const char *file, const char *this_url)
+retrieve_tree (const char *start_url)
{
- char *constr, *filename, *newloc;
- char *canon_this_url = NULL;
- int dt, inl, dash_p_leaf_HTML = FALSE;
- int meta_disallow_follow;
- int this_url_ftp; /* See below the explanation */
- uerr_t err;
- struct urlinfo *rurl;
- urlpos *url_list, *cur_url;
- char *rfile; /* For robots */
- struct urlinfo *u;
-
- assert (this_url != NULL);
- assert (file != NULL);
- /* If quota was exceeded earlier, bail out. */
- if (downloaded_exceeds_quota ())
- return QUOTEXC;
- /* Cache the current URL in the list. */
- if (first_time)
- {
- /* These three operations need to be done only once per Wget
- run. They should probably be at a different location. */
- if (!undesirable_urls)
- undesirable_urls = make_string_hash_table (0);
-
- hash_table_clear (undesirable_urls);
- string_set_add (undesirable_urls, this_url);
- /* Enter this_url to the hash table, in original and "enhanced" form. */
- u = newurl ();
- err = parseurl (this_url, u, 0);
- if (err == URLOK)
- {
- string_set_add (undesirable_urls, u->url);
- if (opt.no_parent)
- base_dir = xstrdup (u->dir); /* Set the base dir. */
- /* Set the canonical this_url to be sent as referer. This
- problem exists only when running the first time. */
- canon_this_url = xstrdup (u->url);
- }
- else
- {
- DEBUGP (("Double yuck! The *base* URL is broken.\n"));
- base_dir = NULL;
- }
- freeurl (u, 1);
- depth = 1;
- robots_host = NULL;
- forbidden = NULL;
- first_time = 0;
- }
- else
- ++depth;
+ uerr_t status = RETROK;
- if (opt.reclevel != INFINITE_RECURSION && depth > opt.reclevel)
- /* We've exceeded the maximum recursion depth specified by the user. */
- {
- if (opt.page_requisites && depth <= opt.reclevel + 1)
- /* When -p is specified, we can do one more partial recursion from the
- "leaf nodes" on the HTML document tree. The recursion is partial in
- that we won't traverse any <A> or <AREA> tags, nor any <LINK> tags
- except for <LINK REL="stylesheet">. */
- dash_p_leaf_HTML = TRUE;
- else
- /* Either -p wasn't specified or it was and we've already gone the one
- extra (pseudo-)level that it affords us, so we need to bail out. */
- {
- DEBUGP (("Recursion depth %d exceeded max. depth %d.\n",
- depth, opt.reclevel));
- --depth;
- return RECLEVELEXC;
- }
- }
+ /* The queue of URLs we need to load. */
+ struct url_queue *queue = url_queue_new ();
- /* Determine whether this_url is an FTP URL. If it is, it means
- that the retrieval is done through proxy. In that case, FTP
- links will be followed by default and recursion will not be
- turned off when following them. */
- this_url_ftp = (urlproto (this_url) == URLFTP);
+ /* The URLs we decided we don't want to load. */
+ struct hash_table *blacklist = make_string_hash_table (0);
- /* Get the URL-s from an HTML file: */
- url_list = get_urls_html (file, canon_this_url ? canon_this_url : this_url,
- dash_p_leaf_HTML, &meta_disallow_follow);
+ /* We'll need various components of this, so better get it over with
+ now. */
+ struct url *start_url_parsed = url_parse (start_url, NULL);
- if (opt.use_robots && meta_disallow_follow)
- {
- /* The META tag says we are not to follow this file. Respect
- that. */
- free_urlpos (url_list);
- url_list = NULL;
- }
+ url_enqueue (queue, xstrdup (start_url), NULL, 0);
+ string_set_add (blacklist, start_url);
- /* Decide what to do with each of the URLs. A URL will be loaded if
- it meets several requirements, discussed later. */
- for (cur_url = url_list; cur_url; cur_url = cur_url->next)
+ while (1)
{
- /* If quota was exceeded earlier, bail out. */
+ int descend = 0;
+ char *url, *referer, *file = NULL;
+ int depth;
+ boolean dash_p_leaf_HTML = FALSE;
+
if (downloaded_exceeds_quota ())
break;
- /* Parse the URL for convenient use in other functions, as well
- as to get the optimized form. It also checks URL integrity. */
- u = newurl ();
- if (parseurl (cur_url->url, u, 0) != URLOK)
- {
- DEBUGP (("Yuck! A bad URL.\n"));
- freeurl (u, 1);
- continue;
- }
- if (u->proto == URLFILE)
- {
- DEBUGP (("Nothing to do with file:// around here.\n"));
- freeurl (u, 1);
- continue;
- }
- assert (u->url != NULL);
- constr = xstrdup (u->url);
-
- /* Several checkings whether a file is acceptable to load:
- 1. check if URL is ftp, and we don't load it
- 2. check for relative links (if relative_only is set)
- 3. check for domain
- 4. check for no-parent
- 5. check for excludes && includes
- 6. check for suffix
- 7. check for same host (if spanhost is unset), with possible
- gethostbyname baggage
- 8. check for robots.txt
-
- Addendum: If the URL is FTP, and it is to be loaded, only the
- domain and suffix settings are "stronger".
-
- Note that .html and (yuck) .htm will get loaded regardless of
- suffix rules (but that is remedied later with unlink) unless
- the depth equals the maximum depth.
-
- More time- and memory- consuming tests should be put later on
- the list. */
-
- /* inl is set if the URL we are working on (constr) is stored in
- undesirable_urls. Using it is crucial to avoid unnecessary
- repeated continuous hits to the hash table. */
- inl = string_set_contains (undesirable_urls, constr);
-
- /* If it is FTP, and FTP is not followed, chuck it out. */
- if (!inl)
- if (u->proto == URLFTP && !opt.follow_ftp && !this_url_ftp)
- {
- DEBUGP (("Uh, it is FTP but i'm not in the mood to follow FTP.\n"));
- string_set_add (undesirable_urls, constr);
- inl = 1;
- }
- /* If it is absolute link and they are not followed, chuck it
- out. */
- if (!inl && u->proto != URLFTP)
- if (opt.relative_only && !cur_url->link_relative_p)
- {
- DEBUGP (("It doesn't really look like a relative link.\n"));
- string_set_add (undesirable_urls, constr);
- inl = 1;
- }
- /* If its domain is not to be accepted/looked-up, chuck it out. */
- if (!inl)
- if (!accept_domain (u))
+
+ if (status == FWRITEERR)
+ break;
+
+ /* Get the next URL from the queue. */
+
+ if (!url_dequeue (queue,
+ (const char **)&url, (const char **)&referer,
+ &depth))
+ break;
+
+ /* And download it. */
+
+ {
+ int dt = 0;
+ char *redirected = NULL;
+ int oldrec = opt.recursive;
+
+ opt.recursive = 0;
+ status = retrieve_url (url, &file, &redirected, NULL, &dt);
+ opt.recursive = oldrec;
+
+ if (redirected)
{
- DEBUGP (("I don't like the smell of that domain.\n"));
- string_set_add (undesirable_urls, constr);
- inl = 1;
+ xfree (url);
+ url = redirected;
}
- /* Check for parent directory. */
- if (!inl && opt.no_parent
- /* If the new URL is FTP and the old was not, ignore
- opt.no_parent. */
- && !(!this_url_ftp && u->proto == URLFTP))
+ if (file && status == RETROK
+ && (dt & RETROKF) && (dt & TEXTHTML))
+ descend = 1;
+ }
+
+ if (descend
+ && depth >= opt.reclevel && opt.reclevel != INFINITE_RECURSION)
{
- /* Check for base_dir first. */
- if (!(base_dir && frontcmp (base_dir, u->dir)))
+ if (opt.page_requisites && depth == opt.reclevel)
+ /* When -p is specified, we can do one more partial
+ recursion from the "leaf nodes" on the HTML document
+ tree. The recursion is partial in that we won't
+ traverse any <A> or <AREA> tags, nor any <LINK> tags
+ except for <LINK REL="stylesheet">. */
+ /* #### This would be the place to implement the TODO
+ entry saying that -p should do two more hops on
+ framesets. */
+ dash_p_leaf_HTML = TRUE;
+ else
{
- /* Failing that, check for parent dir. */
- struct urlinfo *ut = newurl ();
- if (parseurl (this_url, ut, 0) != URLOK)
- DEBUGP (("Double yuck! The *base* URL is broken.\n"));
- else if (!frontcmp (ut->dir, u->dir))
- {
- /* Failing that too, kill the URL. */
- DEBUGP (("Trying to escape parental guidance with no_parent on.\n"));
- string_set_add (undesirable_urls, constr);
- inl = 1;
- }
- freeurl (ut, 1);
+ /* Either -p wasn't specified or it was and we've
+ already gone the one extra (pseudo-)level that it
+ affords us, so we need to bail out. */
+ DEBUGP (("Not descending further; at depth %d, max. %d.\n",
+ depth, opt.reclevel));
+ descend = 0;
}
}
- /* If the file does not match the acceptance list, or is on the
- rejection list, chuck it out. The same goes for the
- directory exclude- and include- lists. */
- if (!inl && (opt.includes || opt.excludes))
+
+ /* If the downloaded document was HTML, parse it and enqueue the
+ links it contains. */
+
+ if (descend)
{
- if (!accdir (u->dir, ALLABS))
+ int meta_disallow_follow = 0;
+ struct urlpos *children = get_urls_html (file, url, dash_p_leaf_HTML,
+ &meta_disallow_follow);
+
+ if (opt.use_robots && meta_disallow_follow)
{
- DEBUGP (("%s (%s) is excluded/not-included.\n", constr, u->dir));
- string_set_add (undesirable_urls, constr);
- inl = 1;
+ free_urlpos (children);
+ children = NULL;
}
- }
- if (!inl)
- {
- char *suf = NULL;
- /* We check for acceptance/rejection rules only for non-HTML
- documents. Since we don't know whether they really are
- HTML, it will be deduced from (an OR-ed list):
-
- 1) u->file is "" (meaning it is a directory)
- 2) suffix exists, AND:
- a) it is "html", OR
- b) it is "htm"
-
- If the file *is* supposed to be HTML, it will *not* be
- subject to acc/rej rules, unless a finite maximum depth has
- been specified and the current depth is the maximum depth. */
- if (!
- (!*u->file
- || (((suf = suffix (constr)) != NULL)
- && ((!strcmp (suf, "html") || !strcmp (suf, "htm"))
- && ((opt.reclevel != INFINITE_RECURSION) &&
- (depth != opt.reclevel))))))
+
+ if (children)
{
- if (!acceptable (u->file))
+ struct urlpos *child = children;
+ struct url *url_parsed = url_parsed = url_parse (url, NULL);
+ assert (url_parsed != NULL);
+
+ for (; child; child = child->next)
{
- DEBUGP (("%s (%s) does not match acc/rej rules.\n",
- constr, u->file));
- string_set_add (undesirable_urls, constr);
- inl = 1;
+ if (descend_url_p (child, url_parsed, depth, start_url_parsed,
+ blacklist))
+ {
+ url_enqueue (queue, xstrdup (child->url->url),
+ xstrdup (url), depth + 1);
+ /* We blacklist the URL we have enqueued, because we
+ don't want to enqueue (and hence download) the
+ same URL twice. */
+ string_set_add (blacklist, child->url->url);
+ }
}
+
+ url_free (url_parsed);
+ free_urlpos (children);
}
- FREE_MAYBE (suf);
}
- /* Optimize the URL (which includes possible DNS lookup) only
- after all other possibilities have been exhausted. */
- if (!inl)
+
+ if (opt.delete_after || (file && !acceptable (file)))
{
- if (!opt.simple_check)
- opt_url (u);
- else
- {
- char *p;
- /* Just lowercase the hostname. */
- for (p = u->host; *p; p++)
- *p = TOLOWER (*p);
- xfree (u->url);
- u->url = str_url (u, 0);
- }
- xfree (constr);
- constr = xstrdup (u->url);
- /* After we have canonicalized the URL, check if we have it
- on the black list. */
- if (string_set_contains (undesirable_urls, constr))
- inl = 1;
- /* This line is bogus. */
- /*string_set_add (undesirable_urls, constr);*/
-
- if (!inl && !((u->proto == URLFTP) && !this_url_ftp))
- if (!opt.spanhost && this_url && !same_host (this_url, constr))
- {
- DEBUGP (("This is not the same hostname as the parent's.\n"));
- string_set_add (undesirable_urls, constr);
- inl = 1;
- }
+ /* Either --delete-after was specified, or we loaded this
+ otherwise rejected (e.g. by -R) HTML file just so we
+ could harvest its hyperlinks -- in either case, delete
+ the local file. */
+ DEBUGP (("Removing file due to %s in recursive_retrieve():\n",
+ opt.delete_after ? "--delete-after" :
+ "recursive rejection criteria"));
+ logprintf (LOG_VERBOSE,
+ (opt.delete_after ? _("Removing %s.\n")
+ : _("Removing %s since it should be rejected.\n")),
+ file);
+ if (unlink (file))
+ logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
}
- /* What about robots.txt? */
- if (!inl && opt.use_robots && u->proto == URLHTTP)
- {
- /* Since Wget knows about only one set of robot rules at a
- time, /robots.txt must be reloaded whenever a new host is
- accessed.
- robots_host holds the host the current `forbid' variable
- is assigned to. */
- if (!robots_host || !same_host (robots_host, u->host))
- {
- FREE_MAYBE (robots_host);
- /* Now make robots_host the new host, no matter what the
- result will be. So if there is no /robots.txt on the
- site, Wget will not retry getting robots all the
- time. */
- robots_host = xstrdup (u->host);
- free_vec (forbidden);
- forbidden = NULL;
- err = retrieve_robots (constr, ROBOTS_FILENAME);
- if (err == ROBOTSOK)
- {
- rurl = robots_url (constr, ROBOTS_FILENAME);
- rfile = url_filename (rurl);
- forbidden = parse_robots (rfile);
- freeurl (rurl, 1);
- xfree (rfile);
- }
- }
+ xfree (url);
+ FREE_MAYBE (referer);
+ FREE_MAYBE (file);
+ }
- /* Now that we have (or don't have) robots, we can check for
- them. */
- if (!robots_match (u, forbidden))
- {
- DEBUGP (("Stuffing %s because %s forbids it.\n", this_url,
- ROBOTS_FILENAME));
- string_set_add (undesirable_urls, constr);
- inl = 1;
- }
+ /* If anything is left of the queue due to a premature exit, free it
+ now. */
+ {
+ char *d1, *d2;
+ int d3;
+ while (url_dequeue (queue, (const char **)&d1, (const char **)&d2, &d3))
+ {
+ xfree (d1);
+ FREE_MAYBE (d2);
+ }
+ }
+ url_queue_delete (queue);
+
+ if (start_url_parsed)
+ url_free (start_url_parsed);
+ string_set_free (blacklist);
+
+ if (downloaded_exceeds_quota ())
+ return QUOTEXC;
+ else if (status == FWRITEERR)
+ return FWRITEERR;
+ else
+ return RETROK;
+}
+
+/* Based on the context provided by retrieve_tree, decide whether a
+ URL is to be descended to. This is only ever called from
+ retrieve_tree, but is in a separate function for clarity. */
+
+static int
+descend_url_p (const struct urlpos *upos, struct url *parent, int depth,
+ struct url *start_url_parsed, struct hash_table *blacklist)
+{
+ struct url *u = upos->url;
+ const char *url = u->url;
+
+ DEBUGP (("Deciding whether to enqueue \"%s\".\n", url));
+
+ if (string_set_contains (blacklist, url))
+ {
+ DEBUGP (("Already on the black list.\n"));
+ goto out;
+ }
+
+ /* Several things to check for:
+ 1. if scheme is not http, and we don't load it
+ 2. check for relative links (if relative_only is set)
+ 3. check for domain
+ 4. check for no-parent
+ 5. check for excludes && includes
+ 6. check for suffix
+ 7. check for same host (if spanhost is unset), with possible
+ gethostbyname baggage
+ 8. check for robots.txt
+
+ Addendum: If the URL is FTP, and it is to be loaded, only the
+ domain and suffix settings are "stronger".
+
+ Note that .html files will get loaded regardless of suffix rules
+ (but that is remedied later with unlink) unless the depth equals
+ the maximum depth.
+
+ More time- and memory- consuming tests should be put later on
+ the list. */
+
+ /* 1. Schemes other than HTTP are normally not recursed into. */
+ if (u->scheme != SCHEME_HTTP
+ && !(u->scheme == SCHEME_FTP && opt.follow_ftp))
+ {
+ DEBUGP (("Not following non-HTTP schemes.\n"));
+ goto blacklist;
+ }
+
+ /* 2. If it is an absolute link and they are not followed, throw it
+ out. */
+ if (u->scheme == SCHEME_HTTP)
+ if (opt.relative_only && !upos->link_relative_p)
+ {
+ DEBUGP (("It doesn't really look like a relative link.\n"));
+ goto blacklist;
+ }
+
+ /* 3. If its domain is not to be accepted/looked-up, chuck it
+ out. */
+ if (!accept_domain (u))
+ {
+ DEBUGP (("The domain was not accepted.\n"));
+ goto blacklist;
+ }
+
+ /* 4. Check for parent directory.
+
+ If we descended to a different host or changed the scheme, ignore
+ opt.no_parent. Also ignore it for -p leaf retrievals. */
+ if (opt.no_parent
+ && u->scheme == parent->scheme
+ && 0 == strcasecmp (u->host, parent->host)
+ && u->port == parent->port)
+ {
+ if (!frontcmp (parent->dir, u->dir))
+ {
+ DEBUGP (("Trying to escape the root directory with no_parent in effect.\n"));
+ goto blacklist;
}
+ }
- filename = NULL;
- /* If it wasn't chucked out, do something with it. */
- if (!inl)
+ /* 5. If the file does not match the acceptance list, or is on the
+ rejection list, chuck it out. The same goes for the directory
+ exclusion and inclusion lists. */
+ if (opt.includes || opt.excludes)
+ {
+ if (!accdir (u->dir, ALLABS))
{
- DEBUGP (("I've decided to load it -> "));
- /* Add it to the list of already-loaded URL-s. */
- string_set_add (undesirable_urls, constr);
- /* Automatically followed FTPs will *not* be downloaded
- recursively. */
- if (u->proto == URLFTP)
- {
- /* Don't you adore side-effects? */
- opt.recursive = 0;
- }
- /* Reset its type. */
- dt = 0;
- /* Retrieve it. */
- retrieve_url (constr, &filename, &newloc,
- canon_this_url ? canon_this_url : this_url, &dt);
- if (u->proto == URLFTP)
- {
- /* Restore... */
- opt.recursive = 1;
- }
- if (newloc)
- {
- xfree (constr);
- constr = newloc;
- }
- /* If there was no error, and the type is text/html, parse
- it recursively. */
- if (dt & TEXTHTML)
+ DEBUGP (("%s (%s) is excluded/not-included.\n", url, u->dir));
+ goto blacklist;
+ }
+ }
+
+ /* 6. */
+ {
+ char *suf = NULL;
+ /* Check for acceptance/rejection rules. We ignore these rules
+ for HTML documents because they might lead to other files which
+ need to be downloaded. Of course, we don't know which
+ documents are HTML before downloading them, so we guess.
+
+ A file is subject to acceptance/rejection rules if:
+
+ * u->file is not "" (i.e. it is not a directory)
+ and either:
+ + there is no file suffix,
+ + or there is a suffix, but is not "html" or "htm",
+ + both:
+ - recursion is not infinite,
+ - and we are at its very end. */
+
+ if (u->file[0] != '\0'
+ && ((suf = suffix (url)) == NULL
+ || (0 != strcmp (suf, "html") && 0 != strcmp (suf, "htm"))
+ || (opt.reclevel == INFINITE_RECURSION && depth >= opt.reclevel)))
+ {
+ if (!acceptable (u->file))
+ {
+ DEBUGP (("%s (%s) does not match acc/rej rules.\n",
+ url, u->file));
+ FREE_MAYBE (suf);
+ goto blacklist;
+ }
+ }
+ FREE_MAYBE (suf);
+ }
+
+ /* 7. */
+ if (u->scheme == parent->scheme)
+ if (!opt.spanhost && 0 != strcasecmp (parent->host, u->host))
+ {
+ DEBUGP (("This is not the same hostname as the parent's (%s and %s).\n",
+ u->host, parent->host));
+ goto blacklist;
+ }
+
+ /* 8. */
+ if (opt.use_robots && u->scheme == SCHEME_HTTP)
+ {
+ struct robot_specs *specs = res_get_specs (u->host, u->port);
+ if (!specs)
+ {
+ char *rfile;
+ if (res_retrieve_file (url, &rfile))
{
- if (dt & RETROKF)
- recursive_retrieve (filename, constr);
+ specs = res_parse_from_file (rfile);
+ xfree (rfile);
}
else
- DEBUGP (("%s is not text/html so we don't chase.\n",
- filename ? filename: "(null)"));
-
- if (opt.delete_after || (filename && !acceptable (filename)))
- /* Either --delete-after was specified, or we loaded this otherwise
- rejected (e.g. by -R) HTML file just so we could harvest its
- hyperlinks -- in either case, delete the local file. */
{
- DEBUGP (("Removing file due to %s in recursive_retrieve():\n",
- opt.delete_after ? "--delete-after" :
- "recursive rejection criteria"));
- logprintf (LOG_VERBOSE,
- (opt.delete_after ? _("Removing %s.\n")
- : _("Removing %s since it should be rejected.\n")),
- filename);
- if (unlink (filename))
- logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
- dt &= ~RETROKF;
+ /* If we cannot get real specs, at least produce
+ dummy ones so that we can register them and stop
+ trying to retrieve them. */
+ specs = res_parse ("", 0);
}
+ res_register_specs (u->host, u->port, specs);
+ }
- /* If everything was OK, and links are to be converted, let's
- store the local filename. */
- if (opt.convert_links && (dt & RETROKF) && (filename != NULL))
- {
- cur_url->convert = CO_CONVERT_TO_RELATIVE;
- cur_url->local_name = xstrdup (filename);
- }
+ /* Now that we have (or don't have) robots.txt specs, we can
+ check what they say. */
+ if (!res_match_path (specs, u->path))
+ {
+ DEBUGP (("Not following %s because robots.txt forbids it.\n", url));
+ goto blacklist;
}
- else
- DEBUGP (("%s already in list, so we don't load.\n", constr));
- /* Free filename and constr. */
- FREE_MAYBE (filename);
- FREE_MAYBE (constr);
- freeurl (u, 1);
- /* Increment the pbuf for the appropriate size. */
}
- if (opt.convert_links && !opt.delete_after)
- /* This is merely the first pass: the links that have been
- successfully downloaded are converted. In the second pass,
- convert_all_links() will also convert those links that have NOT
- been downloaded to their canonical form. */
- convert_links (file, url_list);
- /* Free the linked list of URL-s. */
- free_urlpos (url_list);
- /* Free the canonical this_url. */
- FREE_MAYBE (canon_this_url);
- /* Decrement the recursion depth. */
- --depth;
- if (downloaded_exceeds_quota ())
- return QUOTEXC;
- else
- return RETROK;
+
+ /* The URL has passed all the tests. It can be placed in the
+ download queue. */
+ DEBUGP (("Decided to load it.\n"));
+
+ return 1;
+
+ blacklist:
+ string_set_add (blacklist, url);
+
+ out:
+ DEBUGP (("Decided NOT to load it.\n"));
+
+ return 0;
}
\f
+/* Register that URL has been successfully downloaded to FILE. */
+
void
register_download (const char *url, const char *file)
{
return;
if (!dl_file_url_map)
dl_file_url_map = make_string_hash_table (0);
- hash_table_put (dl_file_url_map, xstrdup (file), xstrdup (url));
if (!dl_url_file_map)
dl_url_file_map = make_string_hash_table (0);
- hash_table_put (dl_url_file_map, xstrdup (url), xstrdup (file));
+
+ if (!hash_table_contains (dl_file_url_map, file))
+ hash_table_put (dl_file_url_map, xstrdup (file), xstrdup (url));
+ if (!hash_table_contains (dl_url_file_map, url))
+ hash_table_put (dl_url_file_map, xstrdup (url), xstrdup (file));
+}
+
+/* Register that FROM has been redirected to TO. This assumes that TO
+ is successfully downloaded and already registered using
+ register_download() above. */
+
+void
+register_redirection (const char *from, const char *to)
+{
+ char *file;
+
+ if (!opt.convert_links)
+ return;
+
+ file = hash_table_get (dl_url_file_map, to);
+ assert (file != NULL);
+ if (!hash_table_contains (dl_url_file_map, from))
+ hash_table_put (dl_url_file_map, xstrdup (from), xstrdup (file));
}
+/* Register that URL corresponds to the HTML file FILE. */
+
void
register_html (const char *url, const char *file)
{
for (html = downloaded_html_files; html; html = html->next)
{
- urlpos *urls, *cur_url;
+ struct urlpos *urls, *cur_url;
char *url;
DEBUGP (("Rescanning %s\n", html->string));
+
/* Determine the URL of the HTML file. get_urls_html will need
it. */
url = hash_table_get (dl_file_url_map, html->string);
DEBUGP (("It should correspond to %s.\n", url));
else
DEBUGP (("I cannot find the corresponding URL.\n"));
+
/* Parse the HTML file... */
urls = get_urls_html (html->string, url, FALSE, NULL);
+
/* We don't respect meta_disallow_follow here because, even if
the file is not followed, we might still want to convert the
links that have been followed from other files. */
+
for (cur_url = urls; cur_url; cur_url = cur_url->next)
{
char *local_name;
+ struct url *u = cur_url->url;
- /* The URL must be in canonical form to be compared. */
- struct urlinfo *u = newurl ();
- uerr_t res = parseurl (cur_url->url, u, 0);
- if (res != URLOK)
- {
- freeurl (u, 1);
- continue;
- }
/* We decide the direction of conversion according to whether
a URL was downloaded. Downloaded URLs will be converted
ABS2REL, whereas non-downloaded will be converted REL2ABS. */
if (local_name)
DEBUGP (("%s marked for conversion, local %s\n",
u->url, local_name));
+
/* Decide on the conversion direction. */
if (local_name)
{
cur_url->convert = CO_CONVERT_TO_COMPLETE;
cur_url->local_name = NULL;
}
- freeurl (u, 1);
}
/* Convert the links in the file. */
convert_links (html->string, urls);
free_urlpos (urls);
}
}
-\f
-/* Robots support. */
-
-/* Construct the robots URL. */
-static struct urlinfo *
-robots_url (const char *url, const char *robots_filename)
-{
- struct urlinfo *u = newurl ();
- uerr_t err;
-
- err = parseurl (url, u, 0);
- assert (err == URLOK && u->proto == URLHTTP);
- xfree (u->file);
- xfree (u->dir);
- xfree (u->url);
- u->dir = xstrdup ("");
- u->file = xstrdup (robots_filename);
- u->url = str_url (u, 0);
- return u;
-}
-
-/* Retrieves the robots_filename from the root server directory, if
- possible. Returns ROBOTSOK if robots were retrieved OK, and
- NOROBOTS if robots could not be retrieved for any reason. */
-static uerr_t
-retrieve_robots (const char *url, const char *robots_filename)
-{
- int dt;
- uerr_t err;
- struct urlinfo *u;
-
- u = robots_url (url, robots_filename);
- logputs (LOG_VERBOSE, _("Loading robots.txt; please ignore errors.\n"));
- err = retrieve_url (u->url, NULL, NULL, NULL, &dt);
- freeurl (u, 1);
- if (err == RETROK)
- return ROBOTSOK;
- else
- return NOROBOTS;
-}
-/* Parse the robots_filename and return the disallowed path components
- in a malloc-ed vector of character pointers.
-
- It should be fully compliant with the syntax as described in the
- file norobots.txt, adopted by the robots mailing list
- (robots@webcrawler.com). */
-static char **
-parse_robots (const char *robots_filename)
+/* Cleanup the data structures associated with recursive retrieving
+ (the variables above). */
+void
+recursive_cleanup (void)
{
- FILE *fp;
- char **entries;
- char *line, *cmd, *str, *p;
- char *base_version, *version;
- int num, i;
- int wget_matched; /* is the part meant for Wget? */
-
- entries = NULL;
-
- num = 0;
- fp = fopen (robots_filename, "rb");
- if (!fp)
- return NULL;
-
- /* Kill version number. */
- if (opt.useragent)
- {
- STRDUP_ALLOCA (base_version, opt.useragent);
- STRDUP_ALLOCA (version, opt.useragent);
- }
- else
- {
- int len = 10 + strlen (version_string);
- base_version = (char *)alloca (len);
- sprintf (base_version, "Wget/%s", version_string);
- version = (char *)alloca (len);
- sprintf (version, "Wget/%s", version_string);
- }
- for (p = version; *p; p++)
- *p = TOLOWER (*p);
- for (p = base_version; *p && *p != '/'; p++)
- *p = TOLOWER (*p);
- *p = '\0';
-
- /* Setting this to 1 means that Wget considers itself under
- restrictions by default, even if the User-Agent field is not
- present. However, if it finds the user-agent set to anything
- other than Wget, the rest will be ignored (up to the following
- User-Agent field). Thus you may have something like:
-
- Disallow: 1
- Disallow: 2
- User-Agent: stupid-robot
- Disallow: 3
- Disallow: 4
- User-Agent: Wget*
- Disallow: 5
- Disallow: 6
- User-Agent: *
- Disallow: 7
-
- In this case the 1, 2, 5, 6 and 7 disallow lines will be
- stored. */
- wget_matched = 1;
- while ((line = read_whole_line (fp)))
+ if (dl_file_url_map)
{
- int len = strlen (line);
- /* Destroy <CR><LF> if present. */
- if (len && line[len - 1] == '\n')
- line[--len] = '\0';
- if (len && line[len - 1] == '\r')
- line[--len] = '\0';
- /* According to specifications, optional space may be at the
- end... */
- DEBUGP (("Line: %s\n", line));
- /* Skip spaces. */
- for (cmd = line; *cmd && ISSPACE (*cmd); cmd++);
- if (!*cmd)
- {
- xfree (line);
- DEBUGP (("(chucked out)\n"));
- continue;
- }
- /* Look for ':'. */
- for (str = cmd; *str && *str != ':'; str++);
- if (!*str)
- {
- xfree (line);
- DEBUGP (("(chucked out)\n"));
- continue;
- }
- /* Zero-terminate the command. */
- *str++ = '\0';
- /* Look for the string beginning... */
- for (; *str && ISSPACE (*str); str++);
- /* Look for comments or trailing spaces and kill them off. */
- for (p = str; *p; p++)
- if (*p && ISSPACE (*p) && ((*(p + 1) == '#') || (*(p + 1) == '\0')))
- {
- /* We have found either a shell-style comment `<sp>+#' or some
- trailing spaces. Now rewind to the beginning of the spaces
- and place '\0' there. */
- while (p > str && ISSPACE (*p))
- --p;
- if (p == str)
- *p = '\0';
- else
- *(p + 1) = '\0';
- break;
- }
- if (!strcasecmp (cmd, "User-agent"))
- {
- int match = 0;
- /* Lowercase the agent string. */
- for (p = str; *p; p++)
- *p = TOLOWER (*p);
- /* If the string is `*', it matches. */
- if (*str == '*' && !*(str + 1))
- match = 1;
- else
- {
- /* If the string contains wildcards, we'll run it through
- fnmatch(). */
- if (has_wildcards_p (str))
- {
- /* If the string contains '/', compare with the full
- version. Else, compare it to base_version. */
- if (strchr (str, '/'))
- match = !fnmatch (str, version, 0);
- else
- match = !fnmatch (str, base_version, 0);
- }
- else /* Substring search */
- {
- if (strstr (version, str))
- match = 1;
- else
- match = 0;
- }
- }
- /* If Wget is not matched, skip all the entries up to the
- next User-agent field. */
- wget_matched = match;
- }
- else if (!wget_matched)
- {
- xfree (line);
- DEBUGP (("(chucking out since it is not applicable for Wget)\n"));
- continue;
- }
- else if (!strcasecmp (cmd, "Disallow"))
- {
- /* If "Disallow" is empty, the robot is welcome. */
- if (!*str)
- {
- free_vec (entries);
- entries = (char **)xmalloc (sizeof (char *));
- *entries = NULL;
- num = 0;
- }
- else
- {
- entries = (char **)xrealloc (entries, (num + 2)* sizeof (char *));
- entries[num] = xstrdup (str);
- entries[++num] = NULL;
- /* Strip trailing spaces, according to specifications. */
- for (i = strlen (str); i >= 0 && ISSPACE (str[i]); i--)
- if (ISSPACE (str[i]))
- str[i] = '\0';
- }
- }
- else
- {
- /* unknown command */
- DEBUGP (("(chucked out)\n"));
- }
- xfree (line);
+ free_keys_and_values (dl_file_url_map);
+ hash_table_destroy (dl_file_url_map);
+ dl_file_url_map = NULL;
}
- fclose (fp);
- return entries;
-}
-
-/* May the URL url be loaded according to disallowing rules stored in
- forbidden? */
-static int
-robots_match (struct urlinfo *u, char **fb)
-{
- int l;
-
- if (!fb)
- return 1;
- DEBUGP (("Matching %s against: ", u->path));
- for (; *fb; fb++)
+ if (dl_url_file_map)
{
- DEBUGP (("%s ", *fb));
- l = strlen (*fb);
- /* If dir is fb, we may not load the file. */
- if (strncmp (u->path, *fb, l) == 0)
- {
- DEBUGP (("matched.\n"));
- return 0; /* Matches, i.e. does not load... */
- }
+ free_keys_and_values (dl_url_file_map);
+ hash_table_destroy (dl_url_file_map);
+ dl_url_file_map = NULL;
}
- DEBUGP (("not matched.\n"));
- return 1;
+ slist_free (downloaded_html_files);
+ downloaded_html_files = NULL;
}
projects / wget / blobdiff